Study finds 29.4 per cent of United States STEM doctoral abstracts carried AI-generated text
16 October 2026
A working paper of the National Bureau of Economic Research finds that 29.4 per cent of abstracts of United States science, technology, engineering and mathematics doctoral dissertations filed from January to May 2026 contained artificial-intelligence-generated text.
Such text first appeared in 2023 at 1.7 per cent of abstracts and rose to 18.8 per cent in 2025.
Mathematical and computer sciences had the highest share at 36 per cent.
About 41 per cent of dissertations by graduates educated in non-English-speaking countries had AI-generated abstracts, against 23 per cent from English-speaking countries.
Among graduates who did their undergraduate studies in the United States, those with AI-generated abstracts were 17 per cent less likely to be employed in academia and 16 per cent more likely to be employed in industry.
The team analysed abstracts from more than 226,000 doctoral graduates between January 2019 and May 2026 using the detection service Pangram, and the paper has not been peer reviewed.
In plain words
NBER National Bureau of Economic Research — A United States research organisation whose working papers circulate findings before peer review.
STEM Science, Technology, Engineering and Mathematics — The group of academic disciplines often treated together in education and workforce policy.